Inter prediction method and apparatus using CPR-based MMVD

Abstract

An image decoding method performed by a decoding apparatus according to the present document comprises the steps of: determining that a merge with motion vector difference (MMVD) mode has been applied to a current block that is a current picture referencing (CPR) coding block referring to a current picture; deriving a base motion information candidate for the current block on the basis of candidate blocks neighboring the current block; generating prediction samples for the current block on the basis of the base motion information candidate; and generating reconstruction samples for the current block on the basis of the prediction samples, wherein the base motion information candidate is derived depending on whether the neighboring candidate blocks are CPR coding blocks.

What is claimed is: 
     
       1. An image decoding method performed by a decoding apparatus, the image decoding method comprising:
 determining that a merge with motion vector difference (MMVD) mode is applied to a current block which is a current picture referencing (CPR) coding block referring to a current picture; 
 determining whether a reference picture list for the current block includes only the current picture or includes one or more reference pictures other than the current picture; 
 deriving a base motion information candidate based on motion information of a neighboring candidate block which is the CPR coding block among neighboring candidate blocks; 
 generating prediction samples for the current block based on the base motion information candidate; and 
 generating reconstructed samples for the current block based on the prediction samples, 
 wherein the base motion information candidate is derived based on the motion information of the neighboring candidate block which is the CPR coding block when the reference picture list includes only the current picture. 
 
     
     
       2. The image decoding method of  claim 1 , wherein a zero vector is configured as the base motion information candidate when all the neighboring candidate blocks are not CPR coding blocks. 
     
     
       3. The image method of  claim 1 , further comprising configuring a motion information candidate list for the current block,
 wherein the motion information candidate list includes the base motion information candidate. 
 
     
     
       4. The image method of  claim 3 , wherein a number of base motion information candidates included in the motion information candidate list is two, and selection information indicating one of the two base motion information candidates is signaled for the current block to which the MMVD mode is applied. 
     
     
       5. The image decoding method of  claim 1 , wherein the deriving of the base motion information candidate for the current block comprises deriving the base motion information candidate based on at least one of motion information of a first neighboring candidate block which is a CPR coding block or motion information of a second neighboring candidate block which is not a CPR coding block among the neighboring candidate blocks. 
     
     
       6. The image method of  claim 5 , further comprising determining whether the reference picture list for the current block includes only the current picture or includes one or more reference pictures other than the current picture,
 wherein, when the reference picture list includes one or more reference pictures other than the current picture, the base motion information candidate is derived based on at least one of the motion information of the first neighboring candidate block which is the CPR coding block or the motion information of the second neighboring candidate block which is not the CPR coding block. 
 
     
     
       7. The image method of  claim 1 , wherein a zero vector is configured as the base motion information candidates when all the neighboring candidate blocks are not available. 
     
     
       8. A image method performed by an encoding apparatus, the image method comprising:
 determining that a merge with motion vector difference (MMVD) mode is applied to a current block which is a current picture referencing (CPR) coding block referring to a current picture; 
 determining whether a reference picture list for the current block includes only the current picture or includes one or more reference pictures other than the current picture; 
 deriving a base motion information candidate based on motion information of a neighboring candidate block which is the CPR coding block among neighboring candidate blocks; 
 generating prediction samples for the current block based on the base motion information candidate; 
 deriving residual samples based on the prediction samples; and 
 encoding image information including information on the residual samples, 
 wherein the base motion information candidate is derived based on the motion information of the neighboring candidate block which is the CPR coding block when the reference picture list includes only the current picture. 
 
     
     
       9. The image method of  claim 8 , wherein a zero vector is configured as the base motion information candidate when all the neighboring candidate blocks are not CPR coding blocks. 
     
     
       10. The image method of  claim 8 , wherein the deriving of the base motion information candidate for the current block comprises deriving the base motion information candidate based on at least one of motion information of a first neighboring candidate block which is a CPR coding block or motion information of a second neighboring candidate block which is not a CPR coding block among the neighboring candidate blocks. 
     
     
       11. The image method of  claim 10 , further comprising determining whether the reference picture list for the current block includes only the current picture or includes one or more reference pictures other than the current picture,
 wherein, when the reference picture list includes one or more reference pictures other than the current picture, the base motion information candidate is derived based on at least one of the motion information of the first neighboring candidate block which is the CPR coding block or the motion information of the second neighboring candidate block which is not the CPR coding block. 
 
     
     
       12. The image method of  claim 8 , wherein a zero vector is configured as the base motion information candidate when all the neighboring candidate blocks are not available.